gui-automation 审计报告.

警告

• Dependency or permission surface needs review
• Permission surface may require sandboxing
• The skill enables arbitrary GUI interaction, shell commands, and host access, but the documentation provides only a one-time consent gate and no broader safety guidance.
• There is no explicit warning about prompt injection or malicious on-screen content steering the agent's clicks, typed text, or shell commands.
• Trajectory sharing uploads session data without cautioning users to review screenshots or avoid exposing secrets and sensitive information.
• Quality score needs review
• Permission surface needs review: shell or command execution, filesystem or document access
• Dependency/runtime risk: command execution surface, external package install surface
• Permission surface: shell or command execution, filesystem or document access

方法

本报告综合公开元数据、AI 审查输出、仓库活跃度、安装就绪度、OpenAgentSkill 事件、质量评分、信任检查和 Agent 安全门槛；它不是完整的源代码安全审计。